Tabyl label

[olivroy]

Description

Addresses #563 (comment)

SImply default to label attribute if it exists when displaying tabyl.

It enables this by default, which is a breaking change? But I wouldn't see why you would not want that if you took care of labelling your variables.

More fiddling required than expected since a lot of base R functions drop attributes subset(), split(), as.character(), etc.

A solution to get the previous behavior is the following (or haven::zap_labels() or similar.

tabyl(mt_label %>% labelled::remove_var_label(), cyl)

Related Issue

Partially addresses #394. Doesn't solve the value labels part, only variable label.

Example

  mt_label <- mtcars
  attr(mt_label$cyl, "label") <- "Number of cyl"
# before 
tabyl(mt_label, cyl)
 cyl  n percent
   4 11 0.34375
   6  7 0.21875
   8 14 0.43750
# this pr 
tabyl(mt_label, cyl)
 Number of cyl  n percent
             4 11 0.34375
             6  7 0.21875
             8 14 0.43750

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (3769ab0) to head (88957ce).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##              main      #575    +/-   ##
==========================================
  Coverage   100.00%   100.00%            
==========================================
  Files           27        27            
  Lines         1274      1401   +127     
==========================================
+ Hits          1274      1401   +127     

Files	Coverage Δ
R/get_dupes.R	100.00% <ø> (ø)
R/print_tabyl.R	100.00% <ø> (ø)
R/tabyl.R	100.00% <100.00%> (ø)

... and 15 files with indirect coverage changes

[sfirke]

Outstanding! Easy to follow, thank you. I just suggested teeny comment changes.

[sfirke]

This is great @olivroy, sorry for delay in review. When we merge this we can see if the Codecov error #576 is fixed 😁

It enables this by default, which is a breaking change? But I wouldn't see why you would not want that if you took care of labelling your variables.

Agree and agree... the next release of janitor should probably be 3.0.0 given that this is a breaking change and that so much time has elapsed. I do expect it could cause pain even if it's an improvement if it breaks existing code.

More fiddling required than expected since a lot of base R functions drop attributes subset(), split(), as.character(), etc.

Yes this is irritating. #527 would at least address the tidyverse functions doing this, I think.

A solution to get the previous behavior is the following (or haven::zap_labels() or similar.

Could you add that workaround to NEWS.md in this PR and move this contribution from "New features" to "Breaking changes"?

[olivroy]

Thanks so much for the review and the kind words! I am very glad my code was easy to follow, but I have to thank you and Bill for that. janitor is one of the first packages I contributed to!

It is a breaking change, but only in display. I don't think many people will pipe much after. And IIRC, the original names are kept in tabyl.

Let me know if you want me to add other documentation, or tests, or do more verification!

[olivroy]

This will work well with #564 as in my experience, names when you read from Excel in particular tend to be more human readable, the ability to keep those and then use them automatically for tabyl() is an improvement in my opinion!